AI for Dubai Real Estate Agents

Dubai agents use ChatGPT and Claude for the evidence side of the job. A seller wants to know what the home is worth. A buyer wants to know whether the asking price is fair. A landlord wants to know what the next tenant will pay. Each question has an answer in Dubai real estate data: the prices DLD recorded for sales and the rents recorded in Ejari contracts.

An assistant on its own cannot see those records. It answers from training data and listing pages, which carry asking prices without dates. The PropertyIndex plugin gives the assistant read-only access to DLD sales, Ejari rental contracts, project analytics and the PIX© market index. The assistant then works like a fast research analyst that can pull any Dubai building's sales on request.

Comparable-sales pack

The recent DLD sales of the same layout in the same building, with date, floor, size, price and price per sqft. It is the evidence behind a listing price.

Pricing range

The median sale and the middle half of sales for that layout over a stated period. It shows the band most buyers actually paid.

Rental evidence

Median annual rent from new Ejari contracts and from renewals in the building. New contracts show what the next tenant pays. Renewals show what sitting tenants pay.

Client summary

A short written paragraph or table built from those figures, stating the period, the number of sales and PropertyIndex as the source.

Across plugin users, the most common request is the sale history of a specific building and layout. Rental contracts, project comparisons and the market index follow. Those requests map directly onto an agent's week.

A comparable-sales pack takes one prompt. The agent asks for the recent ready sales of the same bedroom count in the building. The assistant searches PropertyIndex for the building, queries the DLD sales and lays them out as a table.

Claude

PropertyIndex plugin connected

I have a listing appointment for a ready 2-bed in Creek Rise. Give me the six most recent comparable sales with date, tower, floor, price and price per sqft, then the 12-month median.

PropertyIndex · Searched Creek Rise

PropertyIndex · Queried DLD sales, 2 Bed, ready, Oct 2025 to Sep 2026

Creek Rise recorded 31 ready 2-bedroom sales between October 2025 and September 2026. The six most recent:

Date	Tower, floor	Price	AED/sqft
8 Sep 2026	Podium, G	AED 2,550,000	2,182
1 Sep 2026	Tower 1, 5	AED 2,500,000	2,258
31 Aug 2026	Tower 1, 6	AED 2,550,000	2,300
17 Aug 2026	Tower 2, 18	AED 2,590,000	2,336
5 Aug 2026	Tower 2, 2	AED 2,550,000	2,303
4 Aug 2026	Tower 2, 6	AED 2,600,000	2,348

Across all 31 sales the median was AED 2,550,000, or AED 2,229 per sqft. The middle half sold between AED 2.40M and 2.60M.

The table is short on purpose. Six recent sales of the same layout carry more weight with a seller than a page of community averages. The floor column lets the agent explain a gap between two sales in the same month. The step-by-step method for reading a building's sold prices is covered in how to ask AI what a Dubai property sold for.

1. 1
   Name the building and layout. Give the building name, the number of bedrooms and whether the home is ready or off-plan. The assistant confirms the building it found before it queries.
2. 2
   Ask for the recent sales. Request the six to ten most recent sales with date, floor, size, price and price per sqft. Ask for the 12-month count and median underneath.
3. 3
   Ask for the range. Request the middle half of prices. It shows the band most buyers paid and keeps one outlier from setting the conversation.
4. 4
   Add the rents. Request new contracts and renewals for the same layout. A seller who might let instead of sell needs both numbers.
5. 5
   Write the summary. Ask for a short client paragraph that states the period, the number of sales and PropertyIndex as the source.
6. 6
   Add what you saw. DLD records price, size, floor and date. They do not record view, finish or condition. Your visit supplies those.

A listing price starts from the building's own sales for the same layout. The agent asks the assistant for every bedroom type in the building over the last 12 months. One table then shows where the seller's home sits and what the alternatives sold for.

The 2-bedroom band is narrow. Ten sales between July and September 2026 ran from AED 2.35M to 2.60M, with a median of AED 2.55M. A seller who expects well above AED 2.6M needs a reason the data does not show, such as a high floor, a better view or an upgraded finish.

Financing changes the picture slightly. Of the 31 ready 2-bedroom sales, 13 had a mortgage recorded on the same day, at a median of AED 2.55M. The 18 cash sales had a median of AED 2.40M. An agent can ask the assistant for this split directly, because DLD records whether a sale was financed. The full method for turning comparable sales into a value is set out in AI property valuation in Dubai.

A buyer's “is this overpriced?” has a factual answer once the asking price meets the building's sales. The agent gives the assistant the asking price, the layout and the size. The assistant compares that price with every recent sale of the same layout.

ChatGPT

PropertyIndex plugin connected

My buyer is looking at a ready 2-bed in Creek Rise, about 1,108 sqft, asking AED 2.9M. Is that above recent sales?

PropertyIndex · Searched Creek Rise

PropertyIndex · Queried DLD sales, 2 Bed, ready, Oct 2025 to Sep 2026

Yes. AED 2.9M is above every one of the 31 ready 2-bedroom sales in Creek Rise over the last 12 months.

Measure	Value
Asking price per sqft	about AED 2,615
Median sale per sqft	AED 2,229
Highest sale per sqft	AED 2,461
Highest sale of this layout	AED 2,725,000 (Oct 2025)
Jul to Sep 2026 sales	AED 2.35M to 2.60M

The only sale above AED 2,725,000 was AED 2.8M, for a larger 1,256 sqft layout. The asking price would need a feature the sales do not show, such as a premium view, to make sense.

The agent shows what buyers paid, and the buyer decides how to open.

Landlords ask two questions: what will a new tenant pay, and what are sitting tenants paying? Ejari contracts answer both, because each contract is recorded as new or as a renewal. The agent asks the assistant to split the two.

In Creek Rise, 81 new 2-bedroom contracts starting between October 2025 and September 2026 had a median rent of AED 143,000 a year. The 73 renewals had a median of AED 132,300. The gap tells a landlord with a renewing tenant how far the current rent sits below the market for a new tenant.

The same figures give a gross yield. A median new rent of AED 143,000 against a median ready sale of AED 2,550,000 is a gross yield of about 5.6%. The 1-bedroom layout works out at about 5.8% and the 3-bedroom at about 5.9%. Yield and resale gains for a whole project are covered in AI property investment analysis in Dubai.

An agent may put figures and analysis from the PropertyIndex plugin into client reports with PropertyIndex credited. That covers a CMA for a seller, a short note for a buyer and a rental summary for a landlord. The credit belongs next to the figures, with the period and the number of sales.

A good client summary is short. The agent can ask the assistant for one: “Write a three-sentence summary for a seller of a ready 2-bed in Creek Rise, citing PropertyIndex and the period.” The assistant returns a paragraph such as: Creek Rise recorded 31 ready 2-bedroom sales between October 2025 and September 2026 at a median of AED 2.55M. The middle half sold between AED 2.40M and 2.60M. Source: PropertyIndex, from Dubai Land Department sales.

Two habits keep a report honest. State what DLD records: price, size, floor, date, bedrooms, ready or off-plan, and whether the sale was financed. State what it does not record: view, finish and condition. The agent's own inspection fills that gap, and that is where an agent adds value the data cannot.

The PropertyIndex plugin is licensed for interactive use. A person asks, and the assistant answers in as many steps as it needs. Preparing for a meeting, answering a buyer and writing a client note are all interactive. Work that runs without a person asking belongs on the PIX© API.

Prices, sizes, price per sqft, rents and yields need PIX© Pro. Free accounts see sale counts and dates. Each account receives up to 100,000 records per billing period through the plugin, which is ample for daily agency work. Usage is subject to fair use.

Setup of the plugin takes a few minutes per agent. The steps for both assistants are in how to connect DLD data to ChatGPT and Claude.

Each prompt names a building, a layout and a period, which is what makes the answer specific. Swap in your own building and bedroom count.

Prompts for every other job, from valuations to off-plan checks, are in the ChatGPT and Claude prompt library for Dubai real estate. The wider picture of how assistants answer property questions is in the AI for Dubai real estate overview.